SEO Metrics Explained: What Each Group Tells You Before You Edit the Page
For B2B SaaS product and growth teams: what the 9 SEO metric groups each answer, how to compare them by device and country, and when to edit the page versus repair measurement first.
Brand promotion: this article introduces Guangsuan's SEO measurement guidance.
For a B2B SaaS product and growth team, the weekly argument is familiar: one chart rises, another falls, and someone proposes rewriting the pricing or feature page before lunch. Pause that rewrite. These nine groups of metrics answer different questions. A move in one group is evidence about that stage only, not proof about the whole business. For method definitions, see Guangsuan's 9 SEO metrics guide.
What each group is really counting
Read them in the order a visitor moves: how often a page was offered, how often it was picked, what happened after arrival, whether the page was eligible at all, and what outside signals exist.
- Search visibility and picks. Search Console shows offers and picks inside search. GA4 measures observable activity after people reach your site. A Console pick is therefore not the same thing as a site session.
- Query cover and mean position. A mean position is a weighted summary for the filters you chose, not one live rank every searcher saw. Outside keyword totals rest on their own lists and collection dates, so two tools can differ without either being broken.
- Pick share. Picks divided by offers tells you how often an offered result was chosen in that narrow slice. In a teaching example, 20 clicks from 1,000 impressions is 2%. That line is arithmetic only, not a target for every intent.
- Contacts and ratios. A chat tap, completed form, file download, basket move and paid order are separate outcomes. Test the live event, keep trials, repeats and spam apart from business-confirmed contacts, and name the ratio you used.
- Bounce and engagement under GA4 rules. In GA4, bounce is the share of visits without engagement. Default engagement conditions include time beyond a short threshold, a chosen event, or at least two views, but confirm the actual property setting.
- Focus time and depth against the task. Focus time records time with the page in front. Longer reading can mean careful study or confusion. Finding a specification on one page is not worse than needing four pages.
- Eligibility to appear. Maintain a defined set of canonical pages intended for indexing and prioritise important pages within it. Login areas, parameter copies and material held out by choice stay outside that aim.
- Outside links versus later movement. Total links and referring sites are different counts. A links listing does not confirm present index status for each source. Published, indexed and improved target-page rankings are separate milestones.
- Vendor scores inside one tool only. Moz Domain Authority, Ahrefs DR and other tools use different models and are not interchangeable grades assigned by Google. Compare close peers inside the same tool and window while still reading real pages and outcomes.
Time zones, scope, consent, successful tag execution and canonical attribution can all create differences. Do not force the two totals to agree.
How to read by device and country without fooling yourself
Record language, country, device, branded versus non-branded queries, and the relevant landing pages. Use one finished range, one property and one search type for every export, and leave the unfinished latest day out.
Suppose your SaaS team sells a planning tool and the blended pick share fell after new docs pages started ranking in low positions. Open Pages to find which address moved, then open its query list with offers, picks and position together. Apply the handheld filter, then the country filter, on identical dates. A row with very few offers can swing on one pick, so report offers and picks separately and tag new, continuing and off-task queries instead of sorting by rate alone.
For one important query, read the title and summary actually shown for that country and handheld or desktop view. Ask whether the ranking address serves the same job the query asks for. If the query asks for model, integration, delivery or compatibility but the ranking page only explains ideas, the wrong page type is ranking. Add product or comparison material and an enquiry entry before changing wording. Do not add an unverified certificate, test claim or performance figure to win picks. For whole-site scope across technical repair, content supply and outside resources, see the Guangsuan Google SEO service.
Fix the page now, or fix how you measure first
Guangsuan (光算) groups site work, content and outside resources, with counts, attributes, upkeep and indexing terms checked against the chosen plan. Use this order before you edit:
- Console normal, analytics almost empty: check tags, permission handling, streams and filters. Restore reliable capture before judging demand.
- Offers up, blended share down, picks up: check query and position mix. Compare close slices with the same intent, position band, device and market before rewriting titles.
- Picks up, business-confirmed contacts flat: check intent, market, handheld form and qualification. Walk the product path and read actual enquiry records. Query lines as a rule cannot be tied one by one to a single order.
- Key canonical outside the index: group the indexing report by reason and open samples. Repair only the block proven on the path you checked, such as accidental noindex, blocking, wrong canonical target, unavailable page or absent rendered material, then log a recheck. A crawled-but-not-indexed note alone does not prove thin copy.
Record dates, filters, what changed, who owns it and what stays unverified. Keep absent data marked as absent instead of filling it with another benchmark. Use canonical only for duplicate or highly similar material, after checking the exact robots rule path and user-agent where access is in doubt.
What to prepare before asking for help
Bring facts another person can repeat: site address, target country and language stated separately, affected addresses, one finished export with filters, what changed and when, who can approve code or copy, and how your team qualifies a contact. Note tags and consent handling where analytics sits near empty while Console holds data. This worked check is hypothetical teaching material, not a client case.
FAQ
Do flat offers by themselves call for new copy? No. Flat offers mean the page is seldom offered for the query. Confirm eligibility, entry links and whether the page type fits what results favour first.
Does a lower blended share by itself call for a new title? Not alone. Check whether new low-position queries widened the mix, then compare narrow slices.
Does longer reading time prove the page worked? No. Longer time can mean useful study or confusion. Add the missing condition or next step and check whether the task was completed.
The best product decisions are no longer the loudest in the room — they are the most evidenced.— Obivu Research Note, 2024
See your own customer signals, ranked.
Turn fragmented interviews, tickets, and verbatims into a single feed of validated insights. Full feature access, white-glove onboarding, no credit card.